| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
0 Department of Biology, University of Utah, Salt Lake City, Utah 84112 USA
Received for publication July 16, 1999. Accepted for publication March 9, 2000.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
x) by cavitation corresponded with plant distribution along a moisture gradient. Riparian species were obligate riparian trees (Fraxinus velutina, Populus fremontii, and Salix gooddingii), native shrubs (Baccharis spp.), and an exotic shrub (Tamarix ramosissima). Upland species were evergreen (Juniperus monosperma, Larrea tridentata), drought-deciduous (Ambrosia dumosa, Encelia farinosa, Fouquieria splendens, Cercidium microphyllum), and winter-deciduous (Acacia spp., Prosopis velutina) trees and shrubs. For each species, we measured the "vulnerability curve" of stem xylem, which shows the decrease in hydraulic conductance from cavitation as a function of x and the crit representing the pressure at complete loss of transport. We also measured minimum in situ x(xmin) during the summer drought. Species in desert upland sites were uniformly less vulnerable to cavitation and exhibited lower xmin than riparian species. Values of crit were correlated with minimum x. Safety margins (xmin–crit) tended to increase with decreasing xmin and were small enough that the relatively vulnerable riparian species could not have conducted water at the x experienced in upland habitats (-4 to -10 MPa). Maintenance of positive safety margins in riparian and upland habitats was associated with minimal to no increase in stem cavitation during the summer drought. The absence of less vulnerable species from the riparian zone may have resulted in part from a weak but significant trade-off between decreasing vulnerability to cavitation and conducting efficiency. These data suggest that cavitation vulnerability limits plant distribution by defining maximum drought tolerance across habitats and influencing competitive ability of drought tolerant species in mesic habitats.
Key Words: comparative approach • Sonoran desert vegetation • species distribution • xylem cavitation • xylem conducting efficiency • water relations
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Sperry et al., 1996
; Tyree, 1997
), which is limited in magnitude by cavitation: the sudden change from liquid to vapor phase within normally water-filled xylem conduits. Does this physical limitation on water transport correspond with species distributions with respect to water availability? We addressed this question by surveying cavitation vulnerability of stem xylem across Sonoran desert species adapted to a wide range of water availability from riparian to desert upland habitats.
Water-stress-induced cavitation occurs when the xylem pressure (x) becomes sufficiently negative to overcome the capillary forces of water in the pit membrane pores connecting adjacent air- and water-filled xylem conduits (Zimmermann, 1983
; Crombie, Hipkins, and Milburn, 1985
; Tyree and Sperry, 1989
). The x required for cavitation is determined at least in part by the diameter of the membrane pores; smaller pores allow more negative x before cavitation occurs (Zimmermann, 1983
; Jarbeau, Ewers, and Davis, 1995
). The result of cavitation is a vapor- and air-filled conduit that no longer transports water. The accumulation of such "embolized" conduits reduces xylem hydraulic conductance (kx).
Modeling studies have investigated the consequences of xylem cavitation for limiting the range of possible x allowing water uptake through the soil–plant continuum (Tyree and Sperry, 1988
; Sperry et al., 1998
). For most plants and soils the minimum x allowing water uptake and transport (crit) corresponds to the x causing complete loss of kx. Exceptions may occur when hydraulic failure occurs in the soil rather than the xylem owing to coarse soil texture or limited surface area of absorbing roots. In these cases, the crit is less negative than what would eliminate xylem transport (Sperry et al., 1998
). Thus, the x causing failure of xylem transport represents a conservative (i.e., most negative) estimate for crit. Unless noted, we will use this definition of crit throughout this paper. If the actual x in a plant were ever to drop to crit, all water transport would be eliminated and the plant must either cease gas exchange or suffer the effects of rapid and severe dehydration of its foliage.
Available evidence suggests that the crit limit is of biological and ecological significance, because it appears to correspond with the physiological range of x in plants. In many mesic species, minimum x approaches within a few tenths of a megapascal of crit on a daily basis, suggesting that stomata regulate transpiration to avoid driving x below crit (Tyree and Sperry, 1988
; Sperry, Alder, and Eastlack, 1993
; Sperry and Pockman, 1993
; Hacke and Sauter, 1995
; Saliendra, Sperry, and Comstock, 1995
; Alder, Sperry, and Pockman, 1996
). Furthermore, a review of 37 species from mesic and xeric habitats showed that crit and in situ x were correlated (Sperry, 1995
). The crit of most mesic plants was not sufficiently negative to allow water transport under the low x observed for xeric plants.
Why might safety margins above crit tend to be small? One possibility is that there are disadvantages to having xylem that is overly resistant to cavitation for the prevailing x requirements of a habitat. There may be a trade-off between safety from cavitation during drought and hydraulic efficiency when soil moisture is high (Zimmermann, 1983
; Sobrado, 1993
; Tyree, Davis, and Cochard, 1994
; Alder, Sperry, and Pockman, 1996
). This trade-off could arise if increased resistance to cavitation necessitated smaller diameter conduits with lower conducting capacity or if the smaller diameter pit membrane pores associated with increased resistance to cavitation caused a reduction in kx (e.g., Schulte and Gibson, 1988
). In either case, if the lower hydraulic efficiency of cavitation-resistant plants resulted in slower growth or limited resource acquisition under wet soil conditions that otherwise favored growth, species with less negative crit could have a competitive advantage over those with more negative crit. While montane trees (Sperry et al., 1994
) exhibited no relationship between cavitation vulnerability and hydraulic efficiency, a larger survey (Tyree, Davis, and Cochard, 1994
) found a weak correlation between cavitation vulnerability and conduit diameter, a proxy for hydraulic efficiency. Another potential disadvantage of cavitation-resistant xylem may be that it requires more energy and structural material to produce (Wagner, Ewers, and Davis, 1998
; Hacke, Sperry, Pockman, and Davis, unpublished data).
It is also possible that safety margins above crit are small because there are direct advantages to cavitation under drought conditions. Cavitation may contribute to drought survival by rationing water use to maximize seasonal extraction of soil moisture reserves (Sperry, 1995
). Water released by cavitating conduits can rehydrate drought-stressed leaves (Dixon, Grace, and Tyree, 1984
; Lo Gullo and Salleo, 1992
). Cavitation in easily replaced or short-lived tissues such as roots or leaves (Zimmermann, 1983
; Tyree and Ewers, 1991
; Tyree et al., 1993
; Sperry, 1995
) may protect larger, long-lived structures such as stems during extreme drought. Similarly, cavitation in fine root junctions of desert succulents can decouple them from drying soil (Nobel and Cui, 1992
; North and Nobel, 1992) limiting water loss via the roots during extended drought.
This paper addresses the relationship between cavitation resistance and species distributions within a small geographic area. Specifically, we predicted that we would observe increasingly negative crit in plants at more and more xeric microsites reflecting plant distribution with respect to water availability. This prediction was addressed by surveying cavitation vulnerability across Sonoran desert species adapted to a wide range of water availability—from riparian to desert upland habitats. These habitats represent a moisture gradient from perennially wet soil in the riparian zone to seasonally very dry soil in the uplands. We measured cavitation vulnerability of 15 species and estimated safety margins based on x at the height of the summer drought of 1995. We also compared x, transpiration (E), and in situ cavitation before and after the drought to assess how closely each species approached crit (=x causing complete loss of kx). Water transport properties and anatomy of stem xylem that relate to hydraulic efficiency were measured to see whether a trade-off with cavitation resistance was present. Finally we re-evaluated previously published data on cavitation, habitat, and conducting efficiency in light of our results from Sonoran desert species.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
) located in close proximity to desert habitat. We studied 15 species (from eight families) spanning the spectrum of water availability at the site (Table 1). These species included most of the dominant species in the area and several patterns of vegetative phenology. All individuals used in the experiments occurred within 1 km of each other. Five individuals of each species were labeled for measurements of transpiration and x. Individuals adjacent to these plants were sampled for vulnerability curves.
|
). Mean monthly precipitation (Fig. 1A) over the same period showed an April-June drought and monsoon rains in July and August. Mean monthly precipitation was highly variable as indicated by its coefficient of variation (Fig. 1A). Long-term mean daily maximum temperatures in Tucson (32 km west) exceed 32°C from May to September (Fig. 1B).
|
x)
). The pressure chamber was used to measure C. microphyllum even when no microphylls were present. Leaf x in S. gooddingii was not measured in April 1995, but average data from April 1993 and 1994 are included for comparison (W. T. Pockman and J. S. Sperry, unpublished data). Predawn and midday leaf x in F. splendens were estimated using leaf disc psychrometers (J. R. D. Merrill Specialty Equipment, Logan, Utah, USA) because the absence of a petiole in this species makes pressure chamber measurements difficult. Psychrometers were individually calibrated using NaCl solutions. Samples were collected with a leaf punch at the same time as those of other species and immediately sealed in a psychrometer chamber (Oosterhuis and Wullschleger, 1987
). Psychrometers were placed in an insulated container and held 30–90 min for vapor- and thermal-equilibration (Brown and Chambers, 1987
) before being measured with a microvoltmeter (model PR-55, Wescor Inc, Logan, Utah, USA).
The minimum stem x of the April and July measurements (xmin) was used to estimate the minimum safety margins (xmin - crit) for each species. For most species xmin was the July measurement. To estimate stem x in April, we measured x of bagged leaves or twigs attached directly to stems of the size used in the vulnerability curve measurements (see below). Flow is minimized through bagged organs, promoting equilibration of x with the subtending stem. To eliminate transpiration, leaves or twigs were wrapped in aluminum foil the evening before measurement. Bagged organs were excised and measured with the pressure chamber at the same time that midday leaf x was determined. To facilitate measurement of all species in a short period at the height of the drought in July, stem x was estimated as the same fraction of the difference between predawn and midday leaf x observed in the April measurements. This approach assumes that any changes in hydraulic conductance of the flow path from soil to stem were equally distributed along the flow path.
Transpiration rate
Transpiration at ambient humidity was measured on 3–5 leaves of each of the five labeled plants of each species at the beginning (April) and end (early July) of the summer drought. Measurements were made using a steady-state porometer (model LI-1600, LICOR, Lincoln, Nebraska, USA) equipped with the cylindrical chamber head (model 1600-07). Absolute transpiration was recorded in the field and later corrected by dividing by the actual area of the measured leaves collected following measurement. Although transpiration measured this way may not accurately reflect in situ transpiration (McDermitt, 1990
), we used these data only as a relative measurement.
Transpiring surface area was determined (or estimated for non-laminar structures) using a leaf area meter (LI-COR LI-3100). Areas for species with non-laminar transpiring surfaces were calculated. For twigs of the stem photosynthetic shrubs (C. microphyllum and B. sarothroides) the total area was calculated as the area of any leaves (or microphylls) plus the surface area of the stem. Stem surface area of C. microphyllum was calculated as a conical frustum using caliper measurements of diameter at both ends and the length of the green portion of the twig. Similarly, stem area of square-stemmed B. sarothroides was calculated as four times the average width of the green stem multiplied by its length. For J. monosperma, the area measured by the leaf area meter was taken as the projected area of a cylinder, and the actual surface area was estimated by multiplying the projected area by 
.
Native embolism measurements
Native embolism refers to xylem blockage associated with in situ cavitation. We measured it on stem segments of each species using the hydraulic conductivity method of Sperry et al. (1988)
. In early morning, one branch was collected from each of seven plants in the field and transported to the laboratory inside plastic bags containing a moist paper towel to prevent desiccation. In the laboratory, a stem segment 
100 mm in length and 5 mm in diameter was cut from each branch while it was held underwater to prevent the introduction of additional air emboli. The segment ends were trimmed with a sharp razor blade to eliminate any flow restrictions introduced when stems were cut. Segments were fitted with gaskets and installed in a tubing manifold for measurement of hydraulic conductivity (kh = mass flow rate per pressure gradient). Each stem was measured individually by applying gravity-induced pressure of <10 kPa and measuring the flow through the stem. In these and all other hydraulic measurements, we used an HCl solution adjusted to a pH of 2 and filtered to 0.22 µm to retard microbial growth in the tubing manifold. Tests in woody tissue show no effect of this solution relative to water on hydraulic parameters (Sperry and Saliendra, 1994
). Since this work we have switched to using filtered water as a measuring solution and controlling microbial growth with frequent bleaching of the tubing system. After the initial kh measurement (khi), stems were repeatedly flushed with solution at 100 kPa (to remove embolism) until kh measured between flushes reached a maximum (khf). Embolism was quantified as the percentage khi was below khf:
 |
Vulnerability curves
Vulnerability curves of all species (except A. dumosa) were estimated in stem segments using an air-injection method (Sperry and Saliendra, 1994
). According to the air-seeding mechanism (Zimmermann, 1983
), the negative x required to pull air into a conduit and cause cavitation is equal and opposite to the positive air pressure required to push air into that conduit when x is equal to atmospheric pressure. This equality has been verified for a number of conifers and angiosperm species (Sperry and Tyree, 1990
; Sperry, Perry, and Sullivan, 1991
; Cochard, Cruiziat, and Tyree, 1992
; Sperry and Saliendra, 1994
; Jarbeau, Ewers, and Davis, 1995
; Alder, Sperry, and Pockman, 1996
; Sperry et al., 1996
) including two of the species studied here (S. gooddingii and P. fremontii; Pockman et al., 1995
). This relationship allows vulnerability curves (the decline in kh with decreasing x) to be estimated by measuring the decline in kh as increasing air pressure is applied around the xylem.
Branches of each species were collected in the field (one per plant), misted with water, triple bagged in plastic to prevent desiccation, and transported to the laboratory. Stem segments 0.22–0.55 m in length were cut from each branch underwater, side branches removed, and the ends trimmed with a sharp razor blade. Prior to determination of the vulnerability curve, stem segments were flushed with solution at 100 kPa for 20 min. This insured that the vulnerability curve included all potentially functional xylem rather than just the xylem that was functional at the time of collection. Flushed segments were inserted through a double-ended pressure sleeve with both ends protruding. Rubber compression gaskets, held in place by aluminum end caps, formed an airtight seal so the sleeve pressure could be increased around the stem. One end of the stem was attached to tubing allowing the measurement of kh by applying a hydraulic head and measuring the flow through the segment. Measurements of kh were made gravimetrically by collecting the flow from the stem in tared vials filled with absorbent paper. During all kh measurements, sleeve pressure was held at 0.1 MPa to avoid leakage of solution from any exposed xylem inside the sleeve. Since this work was completed we have found that the 0.1 MPa pressure is usually unnecessary, and less variation results if the sleeve pressure is dropped to atmospheric.
Stem kh was measured after exposure of the segment to progressively higher air pressure in the sleeve. Elevated pressures were held constant for 10–20 min before being decreased to 0.1 MPa and held for 10 min prior to the kh measurement. The kh was not measured at the higher sleeve pressure because rapid airflow through embolized xylem interfered with collection of solution from the exposed segment end. After each increase in air pressure, the percentage embolism was calculated (Eq. 1) where khf was the current segment kh and khi was the kh prior to exposure to pressure above 0.1 MPa. Depending upon species, air pressure was increased in 0.5-, 1-, or 2-MPa increments until the embolism percentage was >95%. The vulnerability curve was plotted as percentage embolism vs. x where x was predicted as the negative of the applied air pressure. To determine whether these vulnerability curves were consistent with field observations, we compared native embolism at xmin with the vulnerability curve for each species.
We used a dehydration method (Sperry, Donnelly, and Tyree, 1988
) for measuring vulnerability curves to corroborate the results of the air injection method (L. tridentata) or where the air injection method yielded inconsistent results because of resinous secretions (E. farinosa). Stems were collected in the field, transported to the laboratory where they were allowed to dehydrate either on the bench-top or, to speed the process, in the sun outside of the laboratory. Stems were then placed in a plastic bag with a moist paper towel for 30 min to allow x to equilibrate throughout the stem before x was measured with a pressure chamber. Percentage embolism was determined on stem segments using the native embolism procedures described above.
Vulnerability curves were used to calculate mean cavitation x, and crit. To calculate mean cavitation x, vulnerability curves were replotted to show the incremental (rather than cumulative) increase in percentage embolism associated with each decrease in x, or "x class." The mean cavitation x for this distribution was determined using the midpoint of each x class. This procedure was performed for the entire vulnerability curve for comparison with measures of xylem conduit diameter and hydraulic efficiency, which were measured on all conduits from a stem cross section (see below). For comparison with xmin, mean cavitation pressure was also determined for vulnerability curves that were scaled to begin with the native embolism level observed in the field (see Results). This procedure yields the mean cavitation pressure of the conduits that were functional at the time of collection. The crit was estimated as x at 100% embolism using a third-order polynomial fitted to the vulnerability curve data. In some cases this procedure resulted in value of crit that was slightly more negative than our pressurization data (e.g., T. ramosissima). The crit of A. dumosa was taken from Mencuccini and Comstock (1997)
for comparison with our measurements of water potential and transpiration in this species.
Hydraulic efficiency parameters
Conduit diameter
Conduit diameter is an important determinant of hydraulic efficiency because flow is proportional to the fourth power of the radius of the conduit (Zimmermann, 1983
). Conduit diameter distributions were measured on all stems used for vulnerability curves. Transverse sections of stems were cut using a sliding microtome or by hand with a razor blade. The lumens of all xylem conduits within a randomly chosen radial sector were traced using a light microscope with a drawing tube attachment and a digitizing tablet (Donsanto Corp., Micro-Plan II, Natick, Massachusetts, USA). To reflect the diameter distribution of all xylem in the stem (vs. one growth ring), a sector was defined as all xylem within a radial sector defined by ray parenchyma. Additional sectors were measured completely until a minimum of 200 conduits was measured from each stem. The digitizing tablet software calculated the maximum diameter and cross-sectional area of each conduit lumen traced. Diameter distributions were then calculated using 10-µm classes based on both the actual percentage of conduits in each class (frequency distribution) and on the estimated percentage of total conductance contributed by each class (hydraulic distribution). The mean of this hydraulically weighted distribution is given by 
r5 divided r4 where r is conduit radius. We refer to this as the "hydraulic mean." Hydraulically weighted distributions and means were used because our measurements of native embolism and vulnerability to cavitation were based on hydraulic conductivity.
For comparison with published data, we also calculated the mean diameter of the conduits that account for 95% of the flow (D95) through stems of each species (Tyree, Davis, and Cochard, 1994
). This was accomplished by sorting the list of measured conduit diameters in descending order, calculating the fourth power of each value and the total r4. Then, the fourth powers were summed again, from largest to smallest, until the value was 95% of total r4 and the mean diameter of this subset of conduits (D95) was calculated.
Specific conductivity (ks)
A direct measure of hydraulic efficiency is the specific conductivity (ks), defined as the kh divided by the conducting area. The ks was calculated by dividing the maximum kh of each stem (from native embolism measurements) by cross-sectional area of the whole stem. Cross sectional areas were measured using a digitizing tablet.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
show vulnerability curves of all species obtained on flushed stems where embolized xylem was initially refilled. The curves show a rather surprising portion of very vulnerable xylem that should be cavitated for typical x measured in the field. Consistent with this observation was the high native embolism of many of these same species (Table 2), most in excess of 50%. Native embolism showed a 1:1 correspondence with the embolism predicted from vulnerability curves for native x (native embolism = 0.949(predicted embolism) + 9.65, r = 0.682, P < 0.01). Dye perfusions of flushed and native stems indicated that most of the highly vulnerable and thus permanently embolized conduits were in older xylem. Thus, the functional xylem was only a small proportion of the total xylem that was refilled during the flushing process. To represent the vulnerability of the functional xylem to cavitation we based the mean cavitation pressure (Fig. 5B) on "scaled" vulnerability curves (not shown) which showed the percentage loss of conductivity with xylem pressure relative to the native embolism values in July.
|
|
|
|
|
x in riparian vs. upland species (Fig. 5A) reflected the extreme differences in water availability between these habitats. Predawn x was greater than -1.2 MPa throughout the study in all riparian species except T. ramosissima. In contrast, among the upland species, April predawn x's were between -0.7 and -3 MPa and August values ranged from -2 to less than -10 MPa. The lowest x among desert upland shrubs were 3–6 times lower than the minimum midday x among riparian species. Two upland species, F. splendens and E. farinosa became deciduous during the summer drought.
Vulnerability to cavitation, crit, and safety margins
Mean cavitation pressure and crit are shown in Fig. 5B. The lower x experienced by desert upland species was associated with greater resistance of these species to cavitation compared to riparian species (Figs. 5, 6). The significant correlation (r = 0.8521, df = 14, P < 0.01) between mean cavitation pressure and xmin (Fig. 6A) indicated that vulnerability to cavitation reflected water availability.
|
crit was also correlated with xmin (Fig. 6B, r = 0.8376, df = 13, P < 0.01). In all species, xmin was above crit, indicating positive safety margins from transport failure. Safety margins ranged from 1.0 to 9.4 MPa and showed a tendency to increase with decreasing xmin (although the slope in Fig. 6B is not significantly different from 1).
The vulnerable xylem of the obligate riparian species (F. velutina, P. fremontii, S. gooddingii; Fig. 2A–C) would not allow them to transport water over the range of x exhibited by most other species in the study (Fig. 6B). Similarly, the xylem of the most xeric species (A. dumosa, E. farinosa, J. monosperma, and L. tridentata; Fig. 4) developed xmin that was low enough to completely cavitate all other species, upland as well as riparian (Fig. 6B). Between these extremes, a larger group of riparian (Baccharis spp., T. ramosissima; Figs. 2D–F) and upland (Acacia sp., C. microphyllum, F. splendens, P. velutina; Figs. 3–4) species exhibited intermediate values of mean cavitation pressure and crit (Figs. 5B, 6B).
Hydraulic efficiency and conduit diameter
Although upland species could easily tolerate the xmin observed in the riparian area, few of these species occurred there. Was this associated with a trade-off between vulnerability to cavitation and hydraulic efficiency? To answer this question, we compared mean cavitation pressure with hydraulic mean diameter (Table 3) and ks (Fig. 5D). Mean cavitation pressure of the nonconiferous species was significantly correlated with mean hydraulic diameter (Fig. 7A; r = 0.6214, df = 12, P < 0.05) and with ks (Fig. 7B; r = 0.7602, df = 12, P < 001) suggesting a trade-off between cavitation vulnerability and conducting efficiency. Correlations with both the hydraulic mean and ks were also significant when J. monosperma was included but, not surprisingly, it had a strong effect on the relationship (Fig. 7). These data indicated that across the species studied increased conduit diameter was generally associated with decreased cavitation resistance.
|
|
Native embolism
Despite the broad range of x experienced by upland species during the summer drought, little or no increase in native embolism occurred between April and July (Table 2). In a majority of the species sampled, percentage embolism remained constant or decreased between April and July (Table 2). Percentage embolism decreased in A. constricta, T. ramosissima, and in both Baccharis species, suggesting that xylem production occurred between April and July and/or that previously embolized vessels had become irreversibly blocked by tyloses during that time. Refilling of embolized vessels with water may also occurred, but seemed unlikely given that the period was one of progressive drought and decreasing x. Dye perfusions showed no evidence of refilling of previous year's xylem. Small but significant increases in native embolism (Table 2) were detected in P. fremontii and A. dumosa. Both species approached crit relatively closely during drought.
Transpiration
In April, the transpiration rate (Fig. 5C) showed no trend across the range of x in riparian and upland habitats (Fig. 5A). However, in July when x of most species was at xmin, transpiration decreased linearly across species with declining predawn x until about -4 MPa (r = 0.8488, df = 10, P < 0.01). Below this x, July transpiration was near zero (Fig. 5C). The riparian species were thus not only the most vulnerable to cavitation and exhibited the smallest safety margins from crit, but they also had the highest transpiration rates in July.
Comparisons within riparian and upland species
Xylem pressure
Besides the differences in x between riparian and upland species, there was also considerable variation within each group. Obligate riparian trees (F. velutiona, P. fremontii, and S. gooddingii) exhibited high predawn x (Fig. 5A), which decreased only slightly between April and July, as did midday x (Fig. 5A). Predawn x in F. velutina was significantly lower than in nearby individuals of P. fremontii (Fig. 5A; Student's t test, t = 7.1759, df = 8, P < 0.01). The mean predawn x in both Baccharis species was lower than nearby riparian trees and decreased between April and July. Predawn x was unexpectedly low in T. ramosissima (Fig. 5A) considering that all individuals were within 5 m of a perennial stream. Perhaps our predawn measurements of this species were influenced by significant nocturnal transpiration or growth induced water uptake.
An even greater range of x was observed among the upland species. Although predawn x in several upland species was greater than -1 MPa in April, the decrease in x between April and July ranged from 0.5 MPa in C. microphyllum to over 9 MPa in A. dumosa. Many of the species at the extremes of this range occurred adjacent to each other (e.g., F. splendens, A. dumosa, and L. tridentata) suggesting that the variation in x was related to differences in rooting depth.
Vulnerability to cavitation and safety margins
The same correlations observed across species between mean cavitation pressure and crit, vs. xmin were also significant when species groups from riparian and upland communities were considered alone. Among riparian species, the native trees (F. velutina, P. fremontii, and S. gooddingii) were the most vulnerable to cavitation (Fig. 2A, B, C) with a crit of -2 to -3 MPa; F. velutina was slightly more resistant than the other two species in accordance with its lower xmin. Interestingly, the nonnative T. ramosissima (Fig. 2D) had a crit of -7 to -8 MPa, 3–4 times more resistant to cavitation than the surrounding native riparian trees. It also had the second largest safety margin in the riparian community (1.9 MPa, smaller only than B. sarothroides). Of the two Baccharis species, B. salicifolia (Fig. 2E), restricted to the floodplain, was more vulnerable to cavitation than B. sarothroides (Fig. 2F), whose range extended to disturbed sites in the desert upland. The vulnerability curves of floodplain and upland populations of B. sarothroides were similar except for a lower crit in the upland population (Fig. 2F).
The upland individuals of the four species of Fabaceae all had a crit of -6 to -8 MPa (Fig. 3). Prosopis velutina (Fig. 3C) sampled in the riparian area within metres of surface water was more vulnerable (crit = -5 MPa) than some similar size individuals in the surrounding desert uplands (crit = -7 to -8 MPa). Three of the six individuals sampled at the upland site had vulnerability curves that were not different from those measured in the riparian area, while the other three were significantly more resistant to cavitation (Fig. 3C, inset).
The other desert upland species exhibited a broad range of vulnerability curves. The drought-deciduous Fouquieria splendens (Fig. 4A) had a crit of -6 MPa, which was more vulnerable to cavitation than both T. ramosissima and B. sarothroides in the floodplain. The two evergreen plant species in the desert upland were the most resistant to cavitation of all the species studied. Juniperus monosperma (Fig. 4C) did not begin to cavitate until pressures below -10 MPa and had a crit of -13 MPa. Though the endpoint of the L. tridentata vulnerability curve was ambiguous (see below), extrapolation suggested a crit of -14 MPa or lower (Fig. 4D). Ambrosia dumosa growing at Organ Pipe Cactus National Monument, Arizona, has been shown to have an crit of approximately -12 MPa (Mencuccini and Comstock, 1997
).
Conduit diameter and hydraulic efficiency
Among the desert upland species, the four representatives of the Fabaceae, together with T. ramoissima and F. velutina in the riparian area, exhibited the largest mean diameter and hydraulic mean diameters of the species studied (Table 3). Despite their large conduits, these species were considerably less vulnerable to cavitation than most of the riparian species discussed previously (Figs. 2, 3). Both actual and hydraulic diameter distributions exhibited a smaller range among the desert shrub species. The mean diameters were quite similar among F. splendens, E. farinosa, and L. tridentata (Table 3) even though their crit ranged from -6 MPa to at least -14 MPa (Fig. 4).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
xmin and crit (Fig. 6B) indicated that most riparian species would be completely cavitated at the x that occurred in the xeric upland habitat. Given that cavitation resistance is to some extent genetically determined (Kavanagh, Bond, and Knowe, 1999
; Kolb and Sperry, 1999a
), the implication is that it has an important influence on the distribution of plants with respect to water availability. Consistent with this is the observation that seedling mortality has been correlated with extensive cavitation (Williams, Davis, and Portwood, 1997
). Cavitation-induced mortality in seedlings and in mature individuals during extreme drought may thus combine to exert an important influence on species distribution.
The relationship that we observed between crit and xmin within a narrow geographic area was reinforced when we combined our data with available values from the literature (Fig. 8A). These data, from plants native to a broad array of mesic and arid, tropical and temperate habitats on three continents, indicated that the plants in this study (Fig. 8A, solid circles) spanned the entire range of x for which vulnerability curves have been measured. The persistence of the relationship across the combined data suggests that vulnerability to cavitation is of general relevance to plant distribution across habitats.
|
crit than mesic species (Figs. 6, 8). There are several factors concerning both xmin and crit that contribute to this trend. Seasonal studies of S. gooddingii, P. velutina, and L. tridentata at the same site suggest that the size of the safety margin is related to the predictability of water availability (Pockman and Sperry, unpublished data). The small safety margins of the obligate riparian species are associated with an abundant perennial water supply and therefore a consistent xmin from year to year. These plants can survive with a small safety margin because they experience predictable water stress. At the other extreme, the evergreen species (J. monosperma, L. tridentata) experience wide variation in water availability and xmin through the seasons. They require large safety margins as insurance against periodically severe drought. Xeric species will also tend to have large safety margins because the xmin measured during a short-term study is likely a significant underestimate of the xmin experienced over the plant lifetime. Many of the study species can live 50–100 yr or longer (Goldberg and Turner, 1986
; Turner, 1990
). The unpredictable climate (Fig. 1A) of the region means that there will be occasional droughts much more severe than we observed. Longer term monitoring of x would tend to reduce safety margins in upland species considerably.
Large safety margins for the upland species in our study may also result from overly negative estimates of crit. Our crit was based on stem xylem. However, as previously mentioned, under xeric conditions hydraulic failure may occur in the soil before it does in the plant (Sperry et al., 1998
) meaning that the actual crit would be less negative than predicted from plant xylem. Even if failure did occur in the plant, several studies indicate that it often occurs in root xylem rather than stem xylem under drought conditions (Alder, Sperry, and Pockman, 1996
; Mencuccini and Comstock, 1997
; Linton, Sperry, and Williams, 1998
; Kolb and Sperry, 1999b
). This is because the root xylem in woody plants is frequently more vulnerable to cavitation than shoots of the same individuals (Alder, Sperry, and Pockman, 1996
; Hacke and Sauter, 1996
; Sperry and Ikeda, 1997
; Linton, Sperry, and Williams, 1998
; Kavanagh, Bond, and Knowe, 1999
). In this way, hydraulic failure may be confined to expendable (replaceable) roots rather than the stem, consistent with the "vulnerability segmentation" concept (Zimmermann, 1983
; Tyree and Ewers, 1991
). The underestimation of crit may help explain why, among the least vulnerable upland species, transpiration decreased to near zero while still maintaining large safety margins from the crit based on stem xylem (Figs. 5, 6).
While the crit data indicate that many of the species in the study could not survive in some of the drier sites (Fig. 6), the explanation for why cavitation-resistant species are excluded from riparian habitats remains an open question. The assumption is that traits conferring drought tolerance, which would include cavitation resistance, are also traits that compromise growth under favorable conditions. The relationship between vulnerability to cavitation and both measures of hydraulic efficiency (Fig. 7) suggested that resistance to cavitation was achieved at the expense of conducting efficiency. When we plotted 50 against D95 (Fig. 9) for our data combined with published data (Tyree, Davis, and Cochard, 1994
) for temperate and tropical trees, shrubs, and vines, we observed a significant correlation between these parameters. This relationship was also significant when the data were averaged by genus or family to compensate for any effects of overrepresentation of phylogenetically related species. While there was no significant relationship among the conifers, 50 was correlated with D95 when all species were considered together. The addition of our data does not substantially change the relationship between 50 and D95 (Fig. 9) reported by Tyree et al. (1994)
. This relationship accounts for 16% of the variation in the data but is too weak to have much predictive value when comparing specific species pairs (e.g., E. farinosa vs. S. gooddingii).
|
Increased cavitation resistance has also been linked to increased wood density in one recent study (Wagner, Ewers, and Davis, 1998
). Such a relationship might be expected if denser wood is required to sustain the compressive forces generated by lower negative pressures, and to minimize air permeability that might nucleate cavitation. Preliminary data from our study species shows strong support for this relationship (Hacke, Sperry, Pockman, and Davis, unpublished data) and suggests that there is an added construction cost to cavitation resistant xylem. Construction of denser wood may also correspond to slower growth rates (Enquist et al., 1999
), further decreasing competitive ability of cavitation resistant species when water is readily available. The picture is thus emerging of several related traits associated with cavitation resistance that together may exclude them from mesic habitats.
The differences in vulnerability between T. ramosissima and native riparian vegetation (Fig. 2) may contribute to its ongoing replacement of S. gooddingii and P. fremontii in riparian areas throughout the American west (Brotherson and Winkel, 1986
; Howe and Knopf, 1991
). The dominant native riparian trees use groundwater to sustain their large canopies and high transpiration rates (Busch, Ingraham, and Smith, 1992
). Along with their profligate water use, such species (Fig. 2; Sperry and Saliendra, 1994
; Tyree et al., 1994
; Alder, Sperry, and Pockman, 1996
) were the most vulnerable to cavitation. Although T. ramosissima also maintained high transpiration rates (Fig. 5C), greater resistance to cavitation (Fig. 2) and lower turgor loss points (Busch and Smith, 1995
) should confer an advantage when its reliable water supply is interrupted or reduced. Interestingly, despite having xylem more resistant to cavitation than the other riparian species, T. ramosissima exhibited similarly high ks (Fig. 5D; Smith et al., 1996
). Cavitation resistance combined with high ks may provide a competitive advantage by allowing it to better tolerate a drop in water table caused by human activity (Smith, Wellington, and Nachlinger, 1991
; Stromberg et al., 1992
) and by its own transpiration (Vitousek, 1990
).
The relationship we observed between crit and xmin (Fig. 6B) reflects not only the distribution of species with respect to water availability but also the association of cavitation vulnerability with plant traits that influence xmin, such as rooting distribution and vegetative phenology. The large differences in crit among species growing next to one another emphasize the importance of this association. For example, F. splendens and P. velutina co-occur with L. tridentata despite having substantially higher crit (Fig. 5B). Although F. splendens is arguably more shallowly rooted than L. tridentata (Cannon, 1911
), the absence of cavitation in the more vulnerable F. splendens (Table 2) suggests that by becoming deciduous it avoids reaching the same xmin as L. tridentata (Fig. 5A). Although the extent of stem water storage in F. splendens has not been determined, this may also play a role in avoiding cavitation or refilling cavitated conduits in this species. The mechanism for this may be similar to shallow-rooted succulents (Nobel and Cui, 1992
; North, Ewers, and Nobel, 1992) where cavitation in roots interrupts hydraulic contact with the soil during drought. Although the vulnerability curve of P. velutina was similar to F. splendens (Figs. 3C, 4A), the deep roots of P. velutina (Stromberg et al., 1992
) mitigate the effects of dry shallow soil allowing the species to remain active without cavitation during drought (Table 2).
The small but significant intraspecific differences that we observed between vulnerability curves of P. velutina and B. sarothroides from different locations within the study site suggest that either vulnerability to cavitation is somewhat plastic in each species or there are distinctly different cavitation genotypes maintained within the population. Previous studies have found a variety of patterns. Acer grandidentatum from different microsites exhibited different root vulnerability curves while stems showed no differences (Alder, Sperry, and Pockman, 1996
). Other studies have shown genetic differences in vulnerability to cavitation among dispersed populations of the same species (Kavanagh, Bond, and Knowe, 1999
; Kolb and Sperry, 1999a
) and little divergence when plants of a common origin were transplanted to different microenvironments (Jackson, Irvine, and Grace, 1995
). Further study will be required to determine whether the differences that we observed are a result of differential success of distinct genotypes in riparian and upland sites or whether these represent evidence of plastic responses of stem xylem.
Our results show that cavitation could play an important role in determining plant distribution within and between broad habitat types. The combined effects of seedling mortality (Williams, Davis, and Portwood, 1997
) and attrition of individuals that become established between climatic extremes may result in the patterns that we observed. Although long-term studies that include physiological data are rare, such efforts and continued study of cavitation during establishment will provide the basis for a mechanistic understanding of the factors that contribute to distribution. A better understanding of the genetic and environmental influence on vulnerability to cavitation is essential if we are to understand the mechanisms resulting in habitat preferences. The role of cavitation and hydraulic efficiency in determining rates of water extraction and growth may also provide useful insight into the mechanisms of competitive exclusion (Eissenstat and Caldwell, 1988
) of drought-adapted species in mesic habitats. Vulnerability to cavitation represents an important adaptation to growth under various water regimes, and it may also prove useful for understanding vegetation change in response to local, regional, and global changes in environmental conditions.
